CN118840802A - Intelligent queuing calling method and system based on machine learning - Google Patents

Abstract

The present invention is applicable to the field of communication control, and provides an intelligent queuing call method and system based on machine learning, the method comprising: estimating the user's cultural level according to the acquired user information, and determining the demand type of the user's pending customer order according to the user information; evaluating and learning the historical service data of the customer service through the server, so as to classify the service skills of the customer service; matching the user's pending customer orders to each server, and predicting the time consumption of each customer order and the total time consumption of all customer orders after matching according to the demand type and service skills based on random forest; assigning a reward degree to each predicted customer order time consumption according to a specified evaluation principle, and adjusting the customer order matching with a reward degree lower than a reward degree threshold; and making a queuing call on the server according to the adjusted customer order matching; the method of the present invention is different from the traditional FCFS queuing mechanism, and provides a more personalized and comfortable service experience.

Description

Intelligent queuing call method and system based on machine learning

Technical Field

The present invention relates to the field of communication control, and in particular to an intelligent queuing calling method and system based on machine learning.

Background Art

With the continuous development and improvement of artificial intelligence, deep learning, machine learning and other technologies, they are helping users improve their production efficiency in more and more industries. Traditional call centers usually adopt a simple first-come, first-served (FCFS) queuing mechanism, which cannot effectively utilize resources and cannot provide personalized service experience for the specific needs of different callers; therefore, it is necessary to provide an intelligent queuing call method and system based on machine learning. The evolving machine learning algorithm makes it possible to achieve more intelligent and personalized call services.

Summary of the invention

The purpose of the present invention is to provide an intelligent queuing call method and system based on machine learning, aiming at solving the problems mentioned in the above background technology.

The first aspect of the present invention is achieved by a machine learning-based intelligent queuing calling method, the method comprising:

Estimate the user's education level based on the acquired user information, and determine the type of demand for the user's pending customer orders based on the user information;

The server evaluates and learns the customer service's historical service data to classify the customer service's service skills; service skills include the types of needs that can be served and the feasibility of serving users with different cultural levels;

Match the user's pending orders to each server, and based on the random forest, predict the time consumption of each order after matching and the total time consumption of all orders according to the demand type and service skills;

According to the specified evaluation principle, the predicted time consumption of each customer order is assigned a reward degree, and the customer order matching with a reward degree lower than the reward degree threshold is adjusted;

Make queue calls on the server side based on the adjusted customer order matching.

Furthermore, the method further comprises:

Collect users' attention to the time spent on customer orders through the server;

According to the attention heat, an additional reward degree is assigned to each predicted customer order time.

Furthermore, the method further comprises:

Reserve redundant servers, which can serve at least as many types of needs as any of the servers in service.

The pending customer orders whose monitored attention heat exceeds the first heat threshold are re-matched to the redundant server.

Furthermore, the method further comprises:

Monitor whether the pending customer orders are marked as expedited. If so, match the pending customer orders marked as expedited to the idle redundant server.

Furthermore, the step of estimating the user's educational level based on the acquired user information and determining the type of demand for the user's pending customer order based on the user information specifically includes:

Based on big data technology, estimate the user's education level according to the acquired user information;

Numerically classify the user's educational level and establish a classification table;

Determine the business types that the server can serve, and determine the demand type of the user's pending customer order based on user information. The demand type corresponds to the business type one by one.

Furthermore, in the step of evaluating and learning the historical service data of the customer service through the server, a reinforcement learning model is used for the evaluation and learning.

Furthermore, before the step of matching the user's pending orders to each service end and predicting the time consumption of each matched order and the total time consumption of all orders based on the demand type and service skills based on the random forest, the method further includes:

Establish a demand type decision tree; establish a service time decision tree;

The demand type decision tree and the service time decision tree are connected to form a random forest.

A second aspect of the present invention is an intelligent queuing call system based on machine learning, used in the method described above, the system comprising:

The user estimation and allocation module is used to estimate the user's educational level based on the acquired user information and determine the type of demand for the user's pending customer order based on the user information;

The model calling module is used to evaluate and learn the customer service historical service data through the server, so as to classify the customer service skills;

The order matching and evaluation module is used to match the user's pending orders to each service end, and based on the random forest, predict the time consumption of each order after matching and the total time consumption of all orders according to the demand type and service skills;

The customer order matching optimization module is used to assign a reward degree to each predicted customer order time consumption according to a specified evaluation principle, and to adjust the customer order matching whose reward degree is lower than the reward degree threshold;

The customer order calling module is used to make queue calls on the server side according to the adjusted customer order matching.

Furthermore, the system further comprises:

The attention acquisition module is used to collect the user's attention to the customer order time consumption through the server;

The customer order optimization module is used to assign additional reward points to the predicted customer order time consumption according to the attention heat.

Furthermore, the system further comprises:

A redundant configuration module is used to reserve a redundant server, and the types of requirements that can be served by the redundant server are not less than any of the servers in service;

A service rematching module, used for rematching the pending customer orders whose monitored attention heat exceeds a first heat threshold to a redundant server;

The urgent customer monitoring and processing module is used to monitor whether the pending customer orders are marked as urgent. If so, the pending customer orders marked as urgent will be matched to the idle redundant server.

The present invention provides an intelligent queuing call method based on machine learning. Compared with the prior art, the present method has the following effects: estimating the user's cultural level based on the acquired user information. Different cultural levels have different understanding abilities of specific business, and customer service may need to provide more detailed and tailored inquiry services; and the professional capabilities of customer service themselves may also be different when dealing with different users (or customers); and the service end evaluates and studies the historical service data of customer service to facilitate the classification of customer service service skills, such as evaluating the types of needs that customer service can serve and the feasibility of serving users with different cultural levels. In this way, the service skills of customer service are linked to the cultural level of the user, reducing the two The difference in matching can better complete the service and provide intelligent and personalized service experience; match the user's pending orders to each server, and based on random forest, predict the time consumption of each order after matching and the total time consumption of all orders according to the demand type and service skills; according to the specified evaluation principle, assign a reward degree to the predicted time consumption of each order, sort them according to the reward degree, and queue calls accordingly, which can ensure that the time consumption of the order is reduced and the total time consumption of all orders is reduced on the basis of achieving personalized matching; it can also adjust the order matching with a reward degree lower than the reward degree threshold, and queue calls on the server according to the adjusted order matching; so as to further reduce the time consumption of each order and improve the processing efficiency and satisfaction of the business.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a flow chart of an intelligent queuing call method based on machine learning provided by an embodiment of the present invention;

FIG2 is a first sub-flow chart of an embodiment of the present invention;

FIG3 is a second sub-flow chart of an embodiment of the present invention;

FIG4 is a third sub-flow chart of an embodiment of the present invention;

FIG5 is a fourth sub-flow chart of an embodiment of the present invention;

FIG6 is a structural block diagram of an intelligent queuing call system based on machine learning provided by an embodiment of the present invention;

FIG7 is a structural block diagram of another intelligent queuing call system based on machine learning provided by an embodiment of the present invention;

FIG8 is a block diagram of the internal structure of a computer device in one embodiment.

DETAILED DESCRIPTION

In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

As shown in FIG1 , in one embodiment, a method for intelligent queuing and calling based on machine learning can be applied to an intelligent pager, which can be composed of a number-taking machine (generating an electronic number or other forms of numbers), a pager, a display device, and a notification system; the method can also be used on a server side and a user side, wherein the server side is usually an office computer and the user side is usually a smart phone; in one scenario, the intelligent pager can be deployed in a bank service outlet, a telecommunications business hall, a shopping mall service center, etc.; taking a bank service outlet as an example, the intelligent pager can be configured as follows:

Number-taking machine: When a user (or customer) arrives at a service outlet, he or she can obtain a queue number (which can be an electronic number) through the number-taking machine. The queue number may contain information such as the service type.

Pager: Customer service (or staff) uses the pager to call the next customer for service;

Display device: such as LED display screen, showing the number currently being served and the list of waiting numbers;

Notification system: notify customers of their upcoming turn through voice broadcasts, text messages, mini-programs, and other methods; multi-channel notifications: in addition to on-site voice broadcasts, customers can also be notified through phone calls, text messages, WeChat, and other methods.

Exemplarily, the method of this embodiment may include steps S101 to S105;

S101: estimating the user's education level based on the acquired user information, and determining the type of demand for the user's pending customer order based on the user information;

For example, user information includes name, gender, age, etc., and even place of residence, ethnicity, and education level; it can be obtained through a number-taking machine, which can be implemented by reading an ID card, an electronic card, etc., or by language input, keyboard input, etc.;

When user information is obtained for the first time, a user profile can usually be established to record the user's service and improve the service level based on the record. If the user information is not obtained for the first time, the user's education level and other data can be directly obtained by calling the record. Generally, the obtained user information may directly contain data related to education level, which also eliminates the estimation step.

Exemplarily, the demand type of the user's pending order is determined according to the user information, and the demand type can be prompted to the user to select or input during the process of obtaining the user information;

For example: the business types of banks can be mainly divided into three categories: asset business, liability business and intermediary business; and asset business includes: loan business: banks provide funds to individuals or enterprises and collect principal and interest in the future; discount business: banks purchase unexpired commercial bills at a price lower than the face value, and recover the face value after maturity; investment business: banks conduct investment activities in financial markets such as bonds and stocks; interbank lending business: banks lend short-term funds to each other. Liability business includes: deposit business: accepting deposits from customers, including demand deposits, time deposits, etc.; borrowing business: banks borrow from the People's Bank of China or other financial institutions; interbank lending: banks borrow short-term funds from other banks or financial institutions; bond issuance business: banks issue bonds to raise funds. Intermediary businesses include: guarantee business: various credit guarantee services provided by banks to customers; agency business: banks act as agents for customers in payment, collection, securities trading, etc.; consulting business: providing consulting services such as financial planning and asset management for enterprises and individuals; payment and collection business: providing transfer, remittance, check cashing and other services; bank card business: issuing credit cards, debit cards, etc.; custody business: providing asset custody services for customers; financial management business: selling financial products and providing investment management services; electronic banking business: providing electronic service channels such as online banking and mobile banking, etc. Among them, the estimation of the user's educational level is mainly based on big data technology and the user's age is estimated.

As shown in FIG. 5 , illustratively, the step of estimating the user's education level based on the acquired user information and determining the type of demand for the user's pending customer order based on the user information specifically includes:

S501: Based on big data technology, estimate the user's education level according to the acquired user information;

Among them, relevant data can be obtained from the server, big data analysis can be performed, and a reference database of age and educational level can be constructed. Based on this database, it is easy to estimate the user's educational level; web crawler technology can also be used to obtain relevant data.

S502: numerically classify the user's educational level and establish a classification table;

For example: establish a cultural grading table for different ages, as shown in Table 1;

Table 1 shows the predicted cultural classification table for different ages

Name age Education grade First 14-16 junior high school Level 1 Second Under 18 high school Level 2 C Less than 22 College or above Level 3 Man 22-40 College or above Level 3 E 40-55 high school Level 2 Self 55 and above Below high school Level 1

S503: Determine the service type that the server can provide, and determine the demand type of the user's pending order based on the user information, where the demand type corresponds to the service type.

Among them, determining the business types that the server can provide is equivalent to determining the business types that each customer service can provide, because the users of each server correspond to the customer service, and the demand type corresponds to the business type one by one, which can improve the efficiency and quality of business handling.

S102: The server evaluates and learns the historical service data of the customer service staff so as to classify the service skills of the customer service staff; wherein the service skills include the types of needs that can be served and the feasibility of serving users with different cultural levels;

Among them, evaluation learning is an algorithm implementation in machine learning, and the specific embodiment of this algorithm can be a reinforcement learning model. The demand type can be generated by machine learning, or edited and classified by senior experts; the feasibility of serving users with different cultural levels is a quantification of the service capabilities of customer service;

The feasibility of customer service for users with different cultural levels indicates the customer service's proficiency in handling demand types and user satisfaction.

S103: Match the pending customer orders of the user to each server, and predict the time consumption of each customer order after matching and the total time consumption of all customer orders based on the demand type and service skills based on random forest;

The user's pending customer order records the demand type, such as specific business types in asset business, liability business and intermediary business;

It is understandable that there may be a situation where multiple user pending orders are matched to one server; in this case, they can be processed according to the FCFS queuing mechanism; the total time consumed for all orders is the sum of the time consumed for each order.

The random forest in this step is a prior art and will not be described in detail here.

In some scenarios, before the step of matching the user's pending orders to each server, and predicting the time consumption of each order after matching and the total time consumption of all orders based on the demand type and service skills based on random forest, the method further includes:

Establish a demand type decision tree; establish a service time decision tree;

The demand type decision tree and the service time decision tree are connected to form a random forest.

S104: assigning a reward degree to each predicted customer order time consumption according to a specified evaluation principle, and adjusting the customer order matching whose reward degree is lower than the reward degree threshold;

The specified evaluation principle may be the shortest time principle, the least number of processing times principle, etc. For the reward level, a reference reward base value may be set according to the average time of each customer order, and then, positive and negative reward values may be assigned according to the time of each customer order and the average time of each customer order to obtain the corresponding reward level; the reward level threshold may also be flexibly set according to actual needs;

For example: the average time taken for a customer order is 18 minutes, and the base reward value is 6 (points); one of the customer orders takes 15 minutes, the assigned positive reward value is 2, and the reward level is 7; the other customer order takes 21 minutes, the assigned negative reward value is -2, and the reward level is 4; the reward level threshold is 5; then, the other customer order is re-matched for adjustment.

S105: Perform a queue call on the server according to the adjusted customer order matching.

In one embodiment, as shown in FIG2 , the method further includes:

S201: Collecting the user's attention to the time spent on a single order through the server;

For example: the electronic number generated by the number-taking machine is displayed on the user's smartphone, and the display interface can be clicked or swiped by the user. The more clicks or swipes there are, the more attention the user pays to time, and the more attention the user pays to the time spent on a customer order. This indirectly indicates how important the customer order is to the user. The number of clicks or swipes is transmitted to the server via the smartphone and processed by the server.

For example, through the video surveillance system, the user's waiting state is collected, and the user's attention to the time consumed by the customer order is judged based on the waiting state; the waiting state can be characterized by the time and number of times the user looks at the display screen (showing a queue).

S202: According to the attention heat, assign an additional reward degree to each predicted customer order consumption time.

The attention heat may be the number of clicks or swipes, or other similar counting methods, which will not be described in detail here.

In the above, according to the attention heat, additional reward degrees are assigned to the predicted time consumption of each customer order; in this way, more urgent customer orders can be weighted so that the customer orders are more likely to be processed with priority.

In an example of this embodiment, as shown in FIG3 , the method further includes:

S301: reserve a redundant server, and the redundant server can serve no fewer types of requirements than any of the active servers;

Of course, the types of demands that can be served by the redundant server can be the same as those of any of the active servers, wherein the types of demands that can be served by the server represent the types of demands that can be served by the customer service corresponding to the server;

In some scenarios, redundant servers can generally be configured for customer service managers or senior customer service staff, etc.

S302: Re-matching the pending customer orders whose monitored attention heat exceeds the first heat threshold to the redundant server.

The first heat threshold can be flexibly set according to actual needs, for example, set to 5, 7 or 10 (times) and so on.

In an example of this embodiment, as shown in FIG4 , the method further includes:

S401: Monitor whether the pending customer orders are marked as expedited, and if so, match the pending customer orders marked as expedited to an idle redundant server.

In an example of this embodiment, in the step of evaluating and learning the historical service data of the customer service through the server, a reinforcement learning model is used for the evaluation and learning.

In another embodiment, as shown in FIG6 , a machine learning-based intelligent queuing calling system is used in the method, and the system includes:

The user estimation and allocation module 100 is used to estimate the user's education level based on the acquired user information and determine the type of demand for the user's pending customer order based on the user information;

The model calling module 200 is used to evaluate and learn the historical service data of the customer service through the server, so as to classify the service skills of the customer service;

The order matching and evaluation module 300 is used to match the user's pending orders to each service end, and predict the time consumption of each order after matching and the total time consumption of all orders based on the demand type and service skills based on random forest;

The customer order matching optimization module 400 is used to assign a reward degree to each predicted customer order time consumption according to a specified evaluation principle, and adjust the customer order matching whose reward degree is lower than the reward degree threshold;

The customer order calling module 500 is used to make a queue call on the server according to the adjusted customer order matching.

In an example of this embodiment, as shown in FIG7 , the system further includes:

The attention acquisition module 600 is used to collect the attention of users on the customer order time consumption through the server;

The customer order optimization module 700 is used to assign additional reward points to each predicted customer order time consumption according to the attention heat.

In an example of this embodiment, as shown in FIG7 , the system further includes:

A redundancy configuration module 800 is used to reserve a redundant server, and the redundant server can serve no less demand types than any server in service;

A service rematching module 900 is used to rematch the monitored pending customer orders whose attention heat exceeds a first heat threshold to a redundant service end;

The urgent customer monitoring and processing module 1000 is used to monitor whether the pending customer orders are marked as urgent, and if so, match the pending customer orders marked as urgent to the idle redundant server.

In an example of this embodiment, the intelligent queuing call system based on machine learning is a computer program that can be executed and implemented by a computer device.

In one embodiment, as shown in FIG8 , a computer device includes a processor and a memory, wherein the memory stores a computer program, and when the processor executes the computer program, steps S101 to S105 of the method are implemented.

The present invention provides an intelligent queuing call method based on machine learning, and based on this method provides an intelligent queuing call system based on machine learning. Compared with the prior art, this method can estimate the user's cultural level based on the acquired user information. Different cultural levels have different understanding abilities of specific business, and customer service may need to provide more detailed inquiry services; and the professional capabilities of customer service themselves may also be different when dealing with different users; and the service end evaluates and studies the historical service data of customer service to facilitate the classification of customer service service skills, such as evaluating the types of needs that customer service can serve and the feasibility of serving users with different cultural levels. In this way, the service skills of customer service are compared with the user. The cultural level of the user is linked to that of the user, and reducing the difference in matching between the two can better complete the service and provide intelligent and personalized service experience; the user's pending customer orders are matched to each service end, and based on random forest, the time consumption of each customer order after matching and the total time consumption of all customer orders are predicted according to the demand type and service skills; according to the specified evaluation principle, a reward degree is assigned to the predicted time consumption of each customer order, and the order is sorted according to the reward degree, and queue calls are made accordingly, which can ensure that the customer order consumption is short on the basis of achieving personalized matching; the customer order matching with a reward degree lower than the reward degree threshold can also be adjusted, and the queue call of the server is made according to the adjusted customer order matching, so as to further reduce the time consumption of each customer order and improve the processing efficiency and satisfaction of the business.

In this embodiment, FIG8 shows an internal structure diagram of a computer device in an embodiment. The computer device includes a processor, a memory, a network interface, an input device and a display screen (or a mobile phone, a laptop computer, an intelligent pager) connected via a system bus. Among them, the memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium of the computer device stores an operating system and may also store a computer program. When the computer program is executed by the processor, the processor can implement an intelligent queuing calling system based on machine learning, specifically, it can implement the steps of each functional module in the system. The internal memory may also store a computer program. When the computer program is executed by the processor, the processor can execute the steps of the intelligent queuing calling method based on machine learning. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse.

Those skilled in the art will understand that the structure shown in FIG. 8 is merely a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

It should be understood that, although each step in the flow chart of each embodiment of the present invention is shown in sequence according to the indication of the arrow, these steps are not necessarily performed in sequence according to the order indicated by the arrow. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and/or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (10)

1. An intelligent queuing call method based on machine learning, characterized in that the method comprises: Estimate the user's education level based on the acquired user information, and determine the type of demand for the user's pending customer orders based on the user information; The server evaluates and learns the customer service's historical service data to facilitate the classification of the customer service skills; Match the user's pending orders to each server, and based on the random forest, predict the time consumption of each order after matching and the total time consumption of all orders according to the demand type and service skills; According to the specified evaluation principle, the predicted time consumption of each customer order is assigned a reward degree, and the customer order matching with a reward degree lower than the reward degree threshold is adjusted; Make queue calls on the server side based on the adjusted customer order matching. 2. The method according to claim 1, characterized in that the method further comprises: Collect users' attention to the time spent on customer orders through the server; According to the attention heat, an additional reward degree is assigned to each predicted customer order time. 3. The method according to claim 2, characterized in that the method further comprises: Reserve redundant servers, which can serve at least as many types of needs as any of the servers in service. The pending customer orders whose monitored attention heat exceeds the first heat threshold are re-matched to the redundant server. 4. The method according to claim 3, characterized in that the method further comprises: Monitor whether the pending customer orders are marked as expedited. If so, match the pending customer orders marked as expedited to the idle redundant server. 5. The method according to claim 1 is characterized in that the step of estimating the user's educational level based on the acquired user information and determining the type of demand for the user's pending customer order based on the user information specifically comprises: Based on big data technology, estimate the user's education level according to the acquired user information; Numerical classification of users' educational level and establishment of a classification table; Determine the business types that the server can serve, and determine the demand type of the user's pending customer order based on user information. The demand type corresponds to the business type one by one. 6. The method according to claim 1 is characterized in that in the step of evaluating and learning the historical service data of the customer service through the server, a reinforcement learning model is used for the evaluation and learning. 7. The method according to claim 1, characterized in that before the step of matching the user's pending orders to each service end and predicting the time consumption of each matched order and the total time consumption of all orders based on the demand type and service skills based on random forest, the method further comprises: Establish a demand type decision tree; establish a service time decision tree; The demand type decision tree and the service time decision tree are connected to form a random forest. 8. An intelligent queuing call system based on machine learning, used in the method according to any one of claims 1 to 7, characterized in that the system comprises: The user estimation and allocation module is used to estimate the user's educational level based on the acquired user information and determine the type of demand for the user's pending customer order based on the user information; The model calling module is used to evaluate and learn the customer service's historical service data through the server, so as to classify the customer service's service skills; The order matching and evaluation module is used to match the user's pending orders to each service end, and based on the random forest, predict the time consumption of each order after matching and the total time consumption of all orders according to the demand type and service skills; The customer order matching optimization module is used to assign a reward degree to each predicted customer order time consumption according to a specified evaluation principle, and to adjust the customer order matching whose reward degree is lower than the reward degree threshold; The customer order calling module is used to make queue calls on the server side according to the adjusted customer order matching. 9. The system according to claim 8, characterized in that the system further comprises: The attention acquisition module is used to collect the user's attention to the customer order time consumption through the server; The customer order optimization module is used to assign additional reward points to the predicted customer order time consumption according to the attention heat. 10. The system according to claim 8, characterized in that the system further comprises: A redundant configuration module is used to reserve a redundant server, and the types of requirements that can be served by the redundant server are not less than any of the servers in service; A service rematching module, used for rematching the pending customer orders whose monitored attention heat exceeds a first heat threshold to a redundant server; The urgent customer monitoring and processing module is used to monitor whether the pending customer orders are marked as urgent. If so, the pending customer orders marked as urgent will be matched to the idle redundant server.